All things related to multicast

["Dale W. Carder" <>]

On Jan 20, 2009, at 8:54 AM, Julian Y. Koh wrote:
> At 08:57 -0500 1/20/2009, Alan Crosswell wrote:
> > The stats from routers will miss all but the first joiner on a subnet
> > with IGMPv2. In the case of, for example, a dorm this could
> > significantly undercount.
>
> That's why we look at the switches, not the routers.  :)

This is the approach I just hacked together in the last few
minutes, too.

Since our edge switches are all the same, I just clogin
"show ip igmp snoop group" to each one.

Surprisingly, collecting it in this manner seems to scale ok.
By running 50 processes in parallel on an 8-way box it can
query ~1000 switch stacks in under a minute.

Right now I'm just collecting the raw output, and I'll have
to parse it later.

Dale

--
Dale W. Carder - Network Engineer
University of Wisconsin / WiscNet
http://net.doit.wisc.edu/~dwcarder